الكلية: كلية الهندسة التكنولوجيّة-قسم هندسة الحاسوب
الجهة البحثية: جامعة البلقاء التطبيقيّة
عنوان البحث المنشور:
Experimental study and computational approach prediction on thermal performance of eutectic salt inside a latent heat storage prototype
حقل البحث: Mechanical Computer
سنة النشر: 2023
ملخص البحث المنشور:
Jordan has made renewable energy a priority in its national policy to achieve renewable energy maturity, particularly in terms of thermal energy storage technologies. The employment of artificial neural network as an efficient tool is the result of a sophisticated mathematical solution of transient non-linear heat equations relevant to thermal storage systems. To forecast the performance of the melting and solidification phenomena of eutectic molten salt inside the heat storage prototype, a computational approach and experimental validation were conducted. In this paper, multilayered perceptron feed-forward artificial neural network with an optimal 2–2-16–2 structure trained by Levenberg-Marquardt algorithm was presented. The stabilization temperature was found to be converged in the region of (174–191) °C for melting mode and around (184–188) °C for solidification mode, which is consistent with the experimental findings. Amount of predicted energy rate during melting/solidification process was around of 1.4/0.77 kW.hr at the inlet temperature of 220 °C whereas it was around 1.9/0.97 kW.hr in case of 250 °C. For testing, the Pearson’s correlation coefficient of the predicted datasets obtained by the artificial neural network was 0.9999. For all situations, the mean square error was roughly 0.00028. The melting and solidification times were approximated more correctly by the artificial neural network model than by the traditional regression method. The data set’s accumulated energy, 220 °C, is likewise excellent, with a performance of around 55 %..
رابط البحث المنشور:
https://www.sciencedirect.com/science/article/abs/pii/S2451904922004127